I often review scientific papers for journals and conferences in computer science. Sometimes when I am asked to do a review I ask my PhD students if they would like to read the paper and discuss reviewing it with me. If they want to join we set a date for discussing the paper, and what to write in a review. The PhD students get course credits from doing these reviews, and I also get a good chance to improve my reviewing skills.
Sometimes the papers we read are very poor, and need very much more work. Those papers are the trickiest ones to review. Often I still write quite a detailed review explaining what can be improved by the paper. In these situations I imagine that the paper is written by a colleague that I really respect, and I am careful with my phrasing of the critique so that it is clear what I mean but put in a way that it sounds polite and nice.
However, once I got such a poor paper that I simply did not write more than:
The ideas in the paper are interesting, but the paper need much more work before it can be published.
One can wonder if this was the correct thing to do???
The paper lacked most parts of a readable paper and it was not possible to understand even what they aimed at doing. A review of such a paper would have meant writing a “How to write a scientific paper for dummies” review. The advise would have been on what to include in an introduction, how to write an abstract etc.
I recommend all seniors to do reviews with their PhD students. It is an interesting way of learning more about papers, and getting a common groups of the area you are doing research in.